- Email: [email protected]
Cognitive behavioral therapy for the treatment of pediatric posttraumatic stress disorder: A review and meta-analysis
J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
Contents lists available at ScienceDirect
Journal of Behavior Therapy and Experimental Psychiatry journal homepage: www.elsevier.com/locate/jbtep
Cognitive behavioral therapy for the treatment of pediatric posttraumatic stress disorder: A review and meta-analysis Joanna Kowalik a, *, Jennifer Weller a, Jacob Venter a, David Drachman b,1 a
University of Arizona College of Medicine Phoenix Campus, Maricopa Integrated Health System/District Medical Group, Desert Vista Behavioral Health Center, 570 West Brown Road, Mesa, AZ 85201, USA b Maricopa Integrated Health System, Maricopa Medical Center, 2601 E. Roosevelt Street, Phoenix, AZ 85008, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 August 2010 Received in revised form 18 January 2011 Accepted 2 February 2011
Background and objectives: There is no clear gold standard treatment for childhood posttraumatic stress disorder (PTSD). An annotated bibliography and meta-analysis were used to examine the efficacy of cognitive behavioral therapy (CBT) in the treatment of pediatric PTSD as measured by outcome data from the Child Behavior Checklist (CBCL). Method: A literature search produced 21 studies; of these, 10 utilized the CBCL but only eight were both 1) randomized; and 2) reported pre- and post-intervention scores. Results: The annotated bibliography revealed efficacy in general of CBT for pediatric PTSD. Using four indices of the CBCL, the meta-analysis identified statistically significant effect sizes for three of the four scales: Total Problems (TP; .327; p ¼ .003), Internalizing (INT; .314; p ¼ .001), and Externalizing (EXT; .192; p ¼ .040). The results for TP and INT were reliable as indicated by the fail-safe N and rank correlation tests. The effect size for the Total Competence (TCOMP; .054; p ¼ .620) index did not reach statistical significance. Limitations: Limitations included methodological inconsistencies across studies and lack of a randomized control group design, yielding few studies for meta-analysis. Conclusions: The efficacy of CBT in the treatment of pediatric PTSD was supported by the annotated bibliography and meta-analysis, contributing to best practices data. CBT addressed internalizing signs and symptoms (as measured by the CBCL) such as anxiety and depression more robustly than it did externalizing symptoms such as aggression and rule-breaking behavior, consistent with its purpose as a therapeutic intervention. Ó 2011 Elsevier Ltd. All rights reserved.
Keywords: PTSD CBT Pediatric Annotated bibliography Meta-analysis
1. Introduction The diagnosis of posttraumatic stress disorder (PTSD) first appeared in the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) in 1980 (American Psychiatric Association, 1980). Posttraumatic Stress Disorder is a complex disorder involving dysregulation of multiple neurobiological systems that affects cognitive, affective, and behavioral domains. Epidemiological studies report a prevalence rate of PTSD in the general adult population ranging between five and 14% (Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995). Trauma is frequently experienced in the United States, with an estimated one-
* Corresponding author. Tel.: þ1 480 344 2026; fax: þ1 480 344 0219. E-mail addresses: [email protected] (J. Kowalik), Jennifer_Weller@ dmgaz.org (J. Weller), [email protected] (J. Venter), David.Drachman@ mihs.org (D. Drachman). 1 Tel.: þ1 602 344 5161; fax: þ1 602 344 1974. 0005-7916/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jbtep.2011.02.002
third of the adult population exposed on an annual basis. Of these individuals, approximately 10e20% will develop the clinical syndrome of PTSD (Solomon & Davidson, 1997). The clinical impression is that, in contrast to the adult population, practitioners treating children and adolescents initially seemed reluctant to apply the diagnosis of PTSD to the pediatric age group. Data regarding trauma exposure as well as the subsequent development of PTSD in youth are more limited compared to adults. According to the National Child Traumatic Stress Network, 25% of children and adolescents experience a traumatic event by the time they reach 16 years of age (Copeland-Linder, 2008). A wide range of trauma rates has been reported in the literature, varying from as low as 16% (Cuffe et al., 1998) to as high as 40% in youth under the age of 18 (Boney-McCoy & Finklehor, 1996; Breslau, Davis, Andreski, & Peterson, 1991; Giaconia et al., 1995; SchwabeStone et al., 1995). The point prevalence of PTSD in youth remains unclear but lifetime prevalence estimates approximate 6% in the pediatric population (Giaconia et al., 1995). This number may be an
406
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
underestimate, given that greater numbers of youth are now receiving the diagnosis as clinicians’ awareness of and comfort with diagnosing PTSD increases. Despite clinicians’ slow recognition of PTSD in younger age groups, it is now accepted as a frequently occurring disorder. Posttraumatic stress disorder in children and adolescents is often severely disabling. Pediatric patients with PTSD present with a multitude of symptoms affecting functioning across different domains. Mueser and Taub (2008) reported a rate of PTSD as high as 20% among youth with severe emotional disorders who were involved in multiple systems of care. The authors also reported that adolescents with PTSD were more likely to engage in a variety of high-risk behaviors including running away from home, self-injury, and delinquency. Adolescents in their study reported higher levels of anxiety and depression and lower levels of optimal functioning in different settings (e.g., at home and at school) than did adolescents without PTSD. Different treatment approaches have been applied to address the symptoms of pediatric PTSD. Currently, outpatient psychotherapy is the preferred initial treatment modality for PTSD, with pharmacotherapy used as an adjunctive intervention (Cohen, 1998). Clinicians use Cognitive Behavioral Therapy (CBT) to address associations between stimuli and conditioned fear responses, the influence of environmental factors on symptom expression, and cognitive and affective regulation; therefore, CBT lends itself to the treatment of symptoms of PTSD. For treatment of sexually abused children, for example, clinicians often use CBT to address sequelae of the trauma including internalizing, externalizing, and sexualized behaviors (MacDonald, Higgins, & Ramchandani, 2006). Taking into account economic factors, CBT provides a focused, time-limited treatment approach to address the effects of trauma and is a cost-effective way of treating a larger number of individuals. Youth with PTSD often require a combination of treatment approaches (e.g., individual, group, and/or family psychotherapy along with pharmacotherapy) potentially in an array of treatment settings, with seamless transitions between levels of care. Treatment decisions are complicated by a lack of empirical data regarding outcomes of particular interventions. For these reasons, it remains difficult to recommend one particular treatment approach over another. Researchers have studied the use of psychotherapy in the pediatric PTSD population, with the majority of studies evaluating the efficacy of CBT approaches (Robertson, Humphreys, & Ray, 2004). Researchers have investigated trauma-focused CBT and found efficacy of this intervention in both individual and group therapy formats for sexually abused youth (Leserman, 2005). Pharmacological studies are fewer in number than studies of CBT, less rigorous in methodology and demonstrate less conclusive findings about efficacy and long-term outcomes (Nikulina et al., 2008). The original purpose of this article was to review the overall efficacy of CBT in the treatment of pediatric PTSD as described in recent literature. This review did not intend to evaluate the efficacy of combinations of treatment interventions or treatment delivered across different clinical settings. It examines immediate rather than long-term outcomes of the intervention. This review explores published research studies and contributes to the understanding and establishment of evidence-based treatment interventions. Once it was observed that the Child Behavior Checklist (CBCL) was the only measure utilized with some consistency across studies of pediatric PTSD, the more focused purpose of the study became to examine the efficacy of CBT in the treatment of pediatric PTSD as measured by outcome data from the CBCL. A quantitative approach that is well-suited for measuring the efficacy of interventions across multiple studies is meta-analysis. In this article, meta-analysis was used to evaluate outcomes from
randomized clinical trials of CBT in which the comparison was to an active control group. For inclusion in a meta-analysis, it is recommended that studies all utilize the same measure of the construct in question in order to maximize comparability among studies (Littell, Corcoran, & Pillai, 2008). Among the studies identified and reviewed in this article, the Child Behavior Checklist (CBCL) was the most commonly used outcome measure for treatment of pediatric PTSD (Achenbach & Edelbrock, 1983). The ability of the CBCL to distinguish between clinical and non-clinical patient samples has been well-established (Kendall, 1994). The CBCL was not designed to assess symptoms of PTSD in particular, and therefore is not considered a measure specific for PTSD phenomenology. It is a descriptive rating measure that assesses parent perceptions of their child’s behavior, adjustment, emotional functioning, and social functioning. Individual symptoms, but not the clinical syndrome of PTSD, are assessed by this measure; however, the CBCL provides composite indices that reflect how PTSD is expressed behaviorally by children and adolescents. The effects of trauma on children are varied and can be expressed in a number of ways. Symptoms can be categorized or conceptualized in ways that are similar to those used by the CBCL; these categories include internalizing, externalizing, and total competence. Some children who are exposed to traumatic experiences may react with internalizing symptoms (e.g., depression, anxiety, and/or somatic complaints as reflected by the Internalizing composite index of the CBCL), some with externalizing symptoms and behaviors (e.g., rulebreaking behaviors and/or aggression as assessed by the Externalizing composite index of the CBCL), and still others may manifest the effects of trauma with features that can be considered part of the “Total Competence” composite index of the CBCL (e.g., social challenges, diminished or limited participation in activities such as sports and hobbies, and school problems). These symptoms and signs are not specific to PTSD per se; however, the effects of trauma can be expressed in these different forms. 2. Method To evaluate the outcomes of CBT treatment studies of pediatric PTSD, the authors conducted a systematic search of data sources for relevant scientific publications. Articles were identified via a search of both Ovid MEDLINE and PsycINFO databases between 1966 and 2010. The following search terms were used: (PTSD OR posttraumatic stress disorder OR sexual abuse) AND (CBT OR cognitive behavioral therapy). The search was subsequently limited to the pediatric population (0e18 years) and the English language. The search identified 21 randomized controlled trials using CBT in the treatment of pediatric PTSD (Celano, Hazzard, Webb, & McCall, 1996; Cohen & Mannarino, 1996, 1998; Cohen, Deblinger, Mannarino, & Steer, 2004; Cohen, Mannarino, & Knudsen, 2005; Cohen, Mannarino, & Staron, 2006; Cohen, Mannarino, Perel, & Staron, 2007; Deblinger, Lippman, & Steer, 1996; Deblinger, Mannarino, Cohen, & Steer, 2006; Deblinger, Stauffer, & Steer, 2001; Deblinger, Steer, & Lippman, 1999; Feather & Ronan, 2006; Giannopoulou, Dikaiakou, & Yule, 2006; Jaberghaderi, Greenwald, Rubin, Zand, and Dalatabadi, 2004; Kazak et al., 2004; King et al., 2000; Kolko, 1996; March, Amaya-Jackson, Murray, & Schulte, 1998; Smith et al., 2007; Stein et al., 2003). Of the 21 studies, two were secondary analyses and therefore were excluded from further review. To be included in the meta-analysis, studies not only had to be randomized, but also had to use the same outcome measure before and immediately at conclusion of the intervention. Only studies comparing CBT to an active control group were included. Among the therapeutic approaches used in the active control groups were supportive unstructured psychotherapy, nondirective supportive treatment, and child-centered therapy. The CBCL was
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
407
Table 1 Annotated bibliography of studies of CBT in pediatric PTSD included in the meta-analysis. 1. Celano et al. (1996) in J Abn Child Psychology Objective Method Participants Interventions
Measures
Outcomes
Comments
To evaluate and compare the efficacy of two short-term individual therapy interventions for sexually abused girls and their non-offending female caretakers Randomized Controlled Trial Participants included 32 sexually abused girls aged 8e13 years and their caretakers from low-income African American families. Participants were randomly assigned to the experimental program (using CBT and metaphoric techniques) or control group (supportive unstructured psychotherapy). Both interventions consisted of eight 1-h sessions. In the experimental group, 30 min were spent with the child and 30 min with the caretaker per session. Two to three sessions included conjoint sessions. In the control group, 40e70% of the session time was with the child, 15e50% was with the caretaker, and 0e25% was with both parties. Out of 56 initial referrals, seven did not meet eligibility criteria, 17 dropped out, and 32 completed treatment. 1. CBCL: INT, EXT, PTSD subscale 2. Children’s Impact of Traumatic Events Scales e Revised (CITES-R) 3. Children’s Global Assessment Scale (CGAS) 4. Parent Reaction to Incest Disclosure Scale (PRIDS) 5. Parental Attribution Scale (PAS) Both treatment programs decreased PTSD symptoms and traumagenic beliefs reflecting self-blame and powerlessness, and increased overall psychosocial functioning. The experimental intervention was more effective than the comparison program in increasing abuse-related caretaker support of the child and in decreasing caretaker self-blame and expectations of undue negative impact of the abuse on the child. Length of the intervention was not specified
2. Cohen and Mannarino (1996) in J Am Acad Child Adolesc Psychiatry Objective Method Participants Interventions Measures Outcomes Comments
To assess treatment outcomes for sexually abused preschool-age children and their parents by comparing the effectiveness of CBT to nondirective supportive treatment (NST) Randomized Controlled Trial Participants included 67 sexually abused preschool children and their parents. Children were randomly assigned to either CBT adapted for sexually abused preschool children (CBT-SAP) or to nondirective supportive therapy. Child Measure: The Preschool Symptom Self-Report (PRESS) Parent Measures: CBCL-Parent Version, Child Sexual Behavior Inventory, Weekly Behavior Report Within-group comparison of pre- and post-treatment outcome measures demonstrated that the NST group did not change significantly with regard to symptomatology, but the CBT group showed significant improvement in symptoms. None
3. Deblinger et al. (1996) in Child Maltreatment Objective
To examine the differential effects of child and non-offending mother participation in CBT designed to treat PTSD and other behavioral and emotional difficulties in school-aged sexually abused children Method Randomized Controlled Trial Participants Participants included 100 families, of which 90 completed the pre-treatment and post-treatment assessment. Of the child group (aged 7e13 years), 83% were female, 17% were male, and 71% had a PTSD diagnosis. Interventions Children were randomly assigned to a control group or to one of three experimental treatment conditions: child only, mother only, or mother and child. Measures 1. Structured Background Interview (SBI) 2. Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS-E) 3. PTSD section of the K-SADS-E 4. CBCL 5. STAIC 6. CDI 7. Parenting Practice Questionnaire (PPQ) Outcomes Results of ANCOVAs showed significant main effects for the CBCL Externalizing scale, CDI, PTSD section of K-SADS-E, and PPQ. Children assigned to experimental treatment (child only or child and parent) reported fewer PTSD symptoms than children in the parent only or control groups. Treated children showed fewer externalizing behaviors. Only six of 38 children in the experimental group met PTSD criteria after the treatment. Comments Heterogeneity of community treatment; low sensitivity of instruments used to assess trauma 4. Cohen and Mannarino (1998) in Child Maltreatment Objective
To evaluate treatment outcomes for recently sexually abused children who received either sexual-specific CBT (SAS-CBT) or nondirective supportive therapy (NST). Method Randomized Controlled Trial Participants Participants included 82 children (7e14 years), but only 49 participants competed treatment and post-treatment evaluation. Interventions Children were randomly assigned to SAS-CBT or NST. Measures 1. CBCL 2. STAIC 3. CDI 4. CSBI Outcomes At post-treatment, children receiving SAS-CBT reported fewer depressive symptoms on the CDI than did children receiving NST. There was a group by time interaction on the CBCL Social Competence scale; the significant pre-treatment difference between groups suggested that the differential improvement of the SAS-CBT group represented regression to the mean. There was no significant difference between treatment groups at post-treatment evaluation, and no significant group by time interaction with regard to sexually inappropriate behaviors. Comments High drop-out rate 5. Deblinger, Steer, and Lippmann (1999) in Child Abuse and Neglect Objective Method
To determine if the 12 session pre- and post-intervention therapeutic gains found by Deblinger et al. (1996) were sustained two years after treatment Randomized Controlled Trial
408
Participants Interventions Measures
Outcomes
Comments
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
The 1996 sample included 100 participating families. Only 90 completed the pre- and post-treatment assessment. Of the children (aged 7e13), 83% were female, 17% were male, and 71% had a PTSD diagnosis. Children were randomly assigned to community comparison (control), a child-only treatment group, a mother-only treatment group, or a mother-and-child treatment group. Participants were assessed three, six, 12, and 24 months after treatment. 1. SBI 2. PTSD section of the K-SADS-E 3. CBCL 4. STAIC 5. CDI 6. PPQ A series of repeated MANCOVAs was conducted, controlling for pre-test scores. Results indicated that, for the measures of psychopathology on which symptoms decreased in the original study (i.e., externalizing problems, depression, and PTSD symptoms), scores on these measures at 3-month, 6-month, 1-year, and 2-year follow-ups were comparable to post-treatment scores. Incomplete data from measures at follow-ups
6. King et al. (2000) in J Am Acad Child Adolesc Psychiatry Objective Method Participants
To evaluate the efficacy of child and caregiver participation in the CBT treatment of sexually abused children with PTSD symptoms Randomized Controlled Trial Participants included 36 sexually abused boys and girls aged 5e17 years with symptoms meeting criteria for PTSD or considered “at high-risk” for PTSD. Only 28 children completed the study. Interventions Intervention groups included two experimental groups and one control group with 12 children in each group. Group 1 received child CBT consisting of 20 50-min sessions targeting abuse-related PTSD symptoms. Group 2 received family CBT consisting of 20 50-min sessions focused on parent training for behavior management and communication. Group 3 was a wait-list control group that received no contact for 24 weeks. An additional 10 children were screened but not included in the study. Eight more children dropped out during treatment (two from the control group, three from Group 1, and three from Group 2). Measures 1. From the Child Behavior Checklist (CBCL), the Internalizing and Externalizing composite scores and the PTSD subscale 2. Anxiety Disorders Interview Schedule (ADIS, Child Version), PTSD section 3. Children’s Depression Inventory (CDI) 4. Fear Thermometer for Sexually Abused Children (FT-SAC) 5. Revised Children’s Manifest Anxiety Scale (RCMAS) 6. Coping Questionnaire for Sexually Abused Children (CQ-SAC) 7. Global Assessment Functioning Scale (GAF) Outcomes Comparing Group 1 and Group 3, children receiving treatment reported significant improvement in PTSD symptoms and fear/anxiety. No difference was detected between child CBT and family CBT treatment groups. Results were still evident at 12-week follow-up. Comments Small study sample 7. Cohen et al. (2004) in J Am Acad Child Adolesc Psychiatry Objective
To examine the differential efficacy of trauma-focused cognitive behavioral therapy (TF-CBT) and child-centered therapy for treating PTSD and related emotional and behavioral problems in children who suffered sexual abuse Method Randomized Controlled Trial Participants Two hundred twenty-nine children (8e14 years) who experienced contact sexual abuse that was confirmed by CPS, law enforcement, or a professional independent evaluator were included in the study. Interventions Children were randomly assigned to the TF-CBT or the child-centered therapy group. Measures 1. Schedule for Affective Disorders and Schizophrenia for School-Aged Children e Present and Lifetime Version (K-SADS-PL) to assess DSM-IV psychiatric disorders 2. Children’s Depression Inventory (CDI) 3. State-Trait Inventory for Children (STAIC) 4. Children’s Attributions and Perceptions Scale (CAPS) Outcomes A series analyses of covariance indicated that children in TF-CBT, compared to those in child-centered therapy, demonstrated significantly more improvement with regard to PTSD, depression, behavior problems, shame, and abuse-related attributions. Comments Lack of non-treatment control group; small diversity of the sample group 8. Cohen et al. (2005) in Child Abuse and Neglect Objective Method Participants Interventions Measures
Outcomes Comments
To measure the durability of improvement in response to two treatments for sexually abused children Randomized Controlled Trial Participants included 82 sexually abused children (8e15 years) and their primary caretakers Children were randomly assigned to either TF-CBT or NST delivered over 12 sessions. 1. CDI 2. Trauma Symptom Checklist for Children (TSCC) 3. STAIC 4. CSBI 5. CBCL The intent-to-treat group indicated significant group time effects in favor of TF-CBT on measures of depression, anxiety, and sexual problems. Among treatment completers, the TF-CBT group showed greater improvement in anxiety, depression, sexual problems, and dissociation. Outcome measures lacked sensitivity for PTSD symptoms; high drop-out rate, especially in the NST group
the most commonly used outcome measure. Nine studies used a variety of scales as outcome measures other than the CBCL, and were excluded. Other outcome measures included but were not limited to the Weekly Behavior Report, the Post-Traumatic Stress Disorder Reaction Index, the State-Trait Anxiety Inventory for Children, the Revised Children’s Manifest Anxiety Scale, the Child Report of Post-Traumatic Symptoms, and the Subjective Units of
Distress Scale. The remaining studies of CBT utilized the CBCL; however, two did not include pre- and post-intervention assessments and were also excluded from the meta-analysis, leaving eight studies total for the meta-analysis. To obtain a comprehensive impression of the studies included in the meta-analysis, the authors compiled an annotated bibliography of each of the eight CBT studies (Table 1).
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
409
In an attempt to enable the CBCL to provide a more specific measure of PTSD symptoms, Wolfe, Gentile, and Wolfe (1989) derived a PTSD scale comprised of a subset of CBCL items. The authors illustrated that the PTSD scale has high discriminant validity, as sexually abused youth scored much higher than did a normative sample. The Wolfe PTSD scale was calculated in only two of the eight studies included in the meta-analysis. In the remaining six studies, scores on the set of items utilized to construct the PTSD scale were not available; therefore, scores for this scale could not be calculated for the remaining studies. The CBCL includes a number of composite indices, of which up to four were utilized to measure outcomes in the eight studies selected for meta-analysis: Total Problems (TP), Internalizing (INT), Externalizing (EXT), and Total Competence (TCOMP). Although the authors were not able to assess a measure specific for PTSD in the meta-analysis, the CBCL-based PTSD scale scores correlate very strongly with the CBCL TP, INT and EXT measures (Ruggiero & McLeer, 2000). Table 2 indicates which CBCL indices were reported in each study included in the meta-analysis. The purpose of the meta-analysis was to calculate an aggregate effect size for the CBT intervention for each of these four CBCL indices. The four metaanalyses were performed with the Comprehensive Meta-Analysis software application, Version 2.2.046 from Biostat, Inc., 2007.
2.3. Publication bias
2.1. Measure of effect size
3.1. Effect sizes
A commonly utilized measure of effect size is Cohen’s d which is simply the standardized difference between the mean outcomes in two experimental groups (Cohen & Yang, 2008). Because several studies included in this meta-analysis had small-sample sizes, Hedges’ g was used to correct for small-sample bias in the Cohen’s d measure and to express the effect size for each study (Hedges,1981). These effect sizes were then aggregated with each effect size weighted using the inverse variance method. This method provides effect sizes with smaller confidence intervals greater weight in the aggregated estimate of effect size (Littell et al., 2008). A negative effect size indicates a lower score on the CBCL index, suggesting improvement in the CBT group compared to the control group.
Fig. 1 and Table 3 report the effect sizes of each CBCL index. For the CBCL TP, INT and EXT indices, effect sizes were statistically significantly in favor of CBT over active control conditions. For the TCOMP index, the average effect size was not statistically significantly different across CBT and control groups. In other words, CBT interventions improved scores on the TP, INT and EXT indices relative to control groups but not on the TCOMP index. For all outcome measures assessed, the I2 statistic indicated substantial homogeneity among the eight studies. To verify the homogeneity assumption, both a fixed effects model (homogeneity assumed) and a random-effects model (heterogeneity assumed) were calculated. The results were exactly the same for TP, EXT, and TCOMP indices, and very similar for the INT index. These findings confirmed the appropriateness of using the fixed effects model in this study. Consequently, the fixed effects model was used to calculate the average effect sizes for each study included in the meta-analysis. (Figs. 2e5)
2.2. Heterogeneity To accurately estimate overall effect sizes for each CBCL index, it is important to know if the differences between calculated effect sizes for each study included in the meta-analysis can be explained by sampling error, or if the variation among effect sizes is due to systematic differences in study characteristics (e.g., such as differences in sample or treatment characteristics). The latter situation is referred to as heterogeneity. For each outcome index included in the meta-analysis, the I2 statistic was used to assess heterogeneity among the studies utilizing that specific measure (Huedo-Medina, Sanchez-Meca, Marin-Martinez, & Botella, 2006).
Studies without significant findings (i.e., null findings) are less likely to be submitted and/or accepted for publication than are studies that find statistically significant results. If unpublished, well-conducted studies with null findings were included in the meta-analysis, the effect size would be reduced. Therefore, it was important to determine if publication bias could have been a factor in the findings. One strategy for detecting publication bias in metaanalyses is to calculate the “fail-safe N.” The fail-safe N is the number of studies with null findings that would have to be added to the meta-analysis to reduce the effect size to a non-significant value (Littell et al., 2008). The rank correlation test can also be used to examine publication bias (Begg & Mazumdar, 1994). When publication bias exists, small studies are more likely to be included when they show a relatively large treatment effect, and more likely to be absent when they show a relatively small treatment effect. The result would then be an inverse correlation between study size and effect size. In this meta-analysis, both the fail-safe N and the rank correlation test were used to assess publication bias. 3. Results
3.2. Publication bias Table 2 reports the fail-safe N for each measure. The fail-safe N for both the TP and INT analyses was four studies. Therefore, it is
Table 2 CBCL Indices reported in studies included in both the annotated bibliography and meta-analysis. Authors Celano et al. (1996) Cohen and Mannarino (1996) Deblinger et al. (1996) Cohen and Mannarino (1998) Deblinger et al. (1999) King et al. (2000) Cohen et al. (2004) Cohen et al. (2005) TOTAL
TP
INT
EXT
X
X X X X
X X 4
X X X 7
X X X X X X X X 8
X
TCOMP
PTSD X
X X X X X 4
2 Fig. 1. Average effect sizes and confidence intervals.
410
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
Table 3 Results of meta-analysis of each CBCL index. Index
Number of studies
Confidence interval
Average effect size
Z statistic
p value
I2
Fail-safe N
Rank correlation
TP INT EXT TCOMP
4 7 8 4
.541 .505 .376 .267
.327 .314 .192 .054
2.297 3.550 2.051 .496
.003 .001 .040 .620
0.0% 20.3% 0.0% 0.0%
4 4 0 0
.167a .571a .179a .167a
a
to to to to
.113 .122 .008 .159
Test for a negative correlation was not significant.
unlikely that publication bias affected the findings for these two CBCL indices. On the other hand, the fail-safe N for the EXT and TCOMP indices was so small (zero) that findings on these two indices should be viewed with caution. None of the rank correlation tests for each of the four CBCL outcome indices showed a significant inverse correlation. These results provided further evidence that publication bias did not significantly affect the findings. 4. Conclusions 4.1. Study strengths and implications The annotated bibliography in Table 1 supported the efficacy in general of CBT for treatment of pediatric PTSD. The meta-analysis of eight randomized trials of CBT, comparing CBT to active control groups and using both pre- and post-intervention assessments, provided strong evidence that CBT is effective in the treatment of childhood PTSD. In particular, Total Problems, Internalizing, and Externalizing indices of the CBCL showed favorable outcomes as reflected by greater effect sizes of the CBT treatment groups versus comparison groups. The findings for the effects of CBT on the Total Competence index of the CBCL were not statistically significant. Participants in the comparison groups in the studies included in this meta-analysis received active treatment as opposed to no treatment at all. This fact strengthens the significance of our conclusion that CBT is efficacious for the treatment of pediatric PTSD. The fail-safe N for both the Total Problems and Internalizing indices was four studies. Therefore, despite the small number of studies available for inclusion in the meta-analysis, the authors can be confident that neither publication bias nor the small number of studies suitable for inclusion significantly affected the findings reflected by these two CBCL indices. In contrast, the fail-safe N for
the Externalizing and Total Competence indices was zero, casting doubt on the validity of effect sizes observed for these dimensions of functioning. Results for the Total Problems and Internalizing indices provided support for the effectiveness of CBT in addressing symptoms specifically measured by these two CBCL indices. Statistically significant positive change occurred in these two indices in response to CBT, and the fail-safe N for these indices indicated that the findings were reliable. Although the Externalizing index showed statistically significant improvement in response to CBT, the fail-safe N of zero indicated that the results should be interpreted with caution. The Total Competence index did not change following CBT and had a fail-safe N of zero; these findings could not be explained satisfactorily and require further exploration. It is possible that CBT interventions for pediatric PTSD do not address externalizing and Total Competence constructs or behaviors as measured on the CBCL to the same degree as they do internalizing symptoms. It would seem that CBT better addresses internalizing symptoms such as anxiety and depression (as the intervention was originally intended to do) than it does externalizing symptoms such as aggression and/or rule-breaking behaviors as they are measured by the CBCL. Externalizing symptoms often lead parents to seek treatment for their children with PTSD, due to significant effects of aggression and rule-breaking behavior on overall social functioning. In addition to CBT, other treatment approaches may be helpful specifically targeting externalizing signs and symptoms that manifest in response to PTSD. 4.2. Study limitations For inclusion in a meta-analysis, studies must utilize similar (or ideally, identical) outcome measures and must use a randomized control group design. The current meta-analysis was limited by
Most Recent
Earliest
Cohen & Mannarino Jan 1996*
Cohen & Mannarino Feb 1998
Each line represents the effect size for the indicated study and all studies occurring before it
Cohen et al. Apr 2004* Cohen et al. 2005*
-1.0
Effect Size
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
Average Effect Size
Favors CBT Group
95% Confidence Interval
Favors Control Group
* Statistically significant
Fig. 2. Cumulative evidence chart: CBCL total problems (TP).
0.8
1.0
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
411
Earliest
Celano et al. Jan 1996 Cohen & Mannarino Jan 1996* Deblinger et al. Nov 1996
Each line represents the effect size for the indicated study and all studies occurring before it
Most Recent
Cohen & Mannarino Feb 1998* King et al. Nov 2000 Cohen et al. Apr 2004* Cohen et al. 2005* -1.0
-0.8
-0.6
-0.4
Effect Size
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Average Effect Size
Favors CBT Group
Favors Control Group
* Statistically significant
95% Confidence Interval
Fig. 3. Cumulative evidence chart: CBCL internalizing (INT).
the diversity of outcome measures utilized across studies, and the absence of randomized control group designs. Some of the outcome measures were designed specifically to measure childhood traumatic experiences, while others tapped emotional and/ or behavioral problems that may be a result of trauma more generally. The CBCL was used most often as an outcome measure among CBT treatment studies; therefore, only studies that utilized this measure were included in the meta-analysis. The use of diverse measures across studies revealed the lack of a “gold standard” instrument for use in the assessment of pediatric PTSD. With respect to use of a randomized control group design, most studies of the treatment of pediatric PTSD using CBT approaches did not utilize such a research design. A further limitation due to the relatively low number of studies that could be included in the meta-analysis was that it limited the generalizability of the findings. A greater number of studies of CBT utilizing the same outcome measure(s), and that were more specifically aimed at assessing PTSD symptoms, would have yielded more robust findings. Inconsistencies across studies in methodology further limited the results of the meta-analysis. Variables such as the type and fidelity of CBT used to address pediatric PTSD (e.g., trauma-focused
CBT, combining traditional CBT with metaphorical techniques, etc.), participant characteristics (e.g., child only receiving intervention versus child and parent receiving intervention), therapist sophistication or training in the intervention(s), and number of sessions and the length of the treatment intervention period all could have affected study findings. Additional clinical characteristics such as age, sex differences, sample sizes, differences in drop-out rates, and severity of symptoms in the study samples could affect the generalizability of findings. 4.3. Future directions Given that this review provided support for the use of CBT to treat symptoms of pediatric PTSD, a next step would be to deconstruct the components of CBT to determine which symptoms of PTSD respond best to particular aspects of this treatment approach. Doing so would allow clinicians to select the ideal combination of treatment components for their pediatric patients who have been exposed to traumatic experiences. Additionally, future studies of the use of CBT in pediatric PTSD would benefit from use of a randomized control group design to enhance scientific merit.
Earliest
Celano et al. Jan 1996 Cohen & Mannarino Jan 1996* Deblinger et al. Nov 1996*
Each line represents the effect size for the indicated study and all studies occurring before it
Most Recent
Cohen & Mannarino Feb 1998 Deblinger et al. 1999 King et al. Nov 2000 Cohen et al. Apr 2004* Cohen et al. 2005* -1.5
Effect Size
-1.0
-0.5
0.0
0.5
Average Effect Size
Favors CBT Group
95% Confidence Interval
Favors Control Group
* Statistically significant
Fig. 4. Cumulative evidence chart: CBCL externalizing (EXT).
1.0
1.5
412
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
Most Recent
Earliest
Cohen & Mannarino Jan 1996
Cohen & Mannarino Feb 1998
Each line represents the effect size for the indicated study and all studies occurring before it
Cohen et al. Apr 2004
Cohen et al. 2005
-1.0
-0.8
-0.6
Effect Size
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Average Effect Size
Favors CBT Group
95% Confidence Interval
Favors Control Group
* Statistically significant
Fig. 5. Cumulative evidence chart: CBCL total competence (TCOMP).
References American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, D.C.: American Psychiatric Press. (DSM-III). Achenbach, T., & Edelbrock, C. (1983). Manual for the child behavior checklist and revised child behavior profile. Univ Vermont/Dept Psychiatry. Begg, C. B., & Mazumdar, M. (1994). Operating characteristics of a rank correlation test for publication bias. Biometrics, 50(4), 1088e1101. Boney-McCoy, S., & Finklehor, D. (1996). Is youth victimization related to trauma symptoms and depression after controlling for prior symptoms and family relationships? A longitudinal, prospective study. Journal of Consulting and Clinical Psychology, 64(6), 1406e1416. Breslau, N., Davis, G. C., Andreski, P., & Peterson, E. (1991). Traumatic events and postrraumatic stress disorder in an urban population of young adults. Archives of General Psychiatry, 48(3), 216e222. Celano, M., Hazzard, A., Webb, C., & McCall, C. (1996). Treatment of traumagenic beliefs among sexually abused girls and their mothers: an evaluation study. Journal of Abnormal Child Psychology, 24(1), 1e17. Cohen, J. A. (1998). Practice parameters for the assessment and treatment of children and adolescents with posttraumatic stress disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 37(10 Suppl.), 4se26s. Cohen, J. A., Deblinger, E., Mannarino, A. P., & Steer, R. (2004). A multisite, randomized controlled trial for children with sexual abuse-related PTSD symptoms. Journal of the American Academy of Child and Adolescent Psychiatry, 43(4), 393e402. Cohen, J. A., & Mannarino, A. P. (1996). A treatment outcome study for sexually abused preschool children: initial findings. Journal of the American Academy of Child and Adolescent Psychiatry, 35(1), 42e50. Cohen, J. A., & Mannarino, A. P. (1998). Interventions for sexually abused children: initial treatment outcome findings. Child Maltreatment, 3(1), 17e26. Cohen, J. A., Mannarino, A. P., & Knudsen, K. (2005). Treating sexually abused children: 1 year follow-up of a randomized controlled trial. Child Abuse and Neglect, 29(2), 103e105. Cohen, J. A., Mannarino, A. P., Perel, J. M., & Staron, V. (2007). A pilot randomized controlled trial of combined trauma-focused CBT and sertraline for childhood PTSD symptoms. Journal of the American Academy of Child and Adolescent Psychiatry, 46(7), 811e819. Cohen, J. A., Mannarino, A. P., & Staron, V. R. (2006). A pilot study of modified cognitive-behavioral therapy for childhood traumatic grief (CBT-CTG). Journal of the American Academy of Child and Adolescent Psychiatry, 45(12), 1465e1473. Cohen, J., & Yang, F. (2008). Progress in applied mathematical modeling. Nova Science Pub Inc. Copeland-Linder, N. (2008). Posttraumatic stress disorder. Pediatrics in Review, 29, 103e104. Cuffe, S. P., Addy, C. L., Garrison, C. Z., Waller, J. L., Jackson, K. L., McKeown, R. E., et al. (1998). Prevalence of PTSD in a community sample of older adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 37(2), 147e154. Deblinger, E., Lippman, J., & Steer, R. (1996). Sexually abused children suffering posttraumatic stress symptoms: initial treatment outcome findings. Child Maltreatment, 1(4), 310e321. Deblinger, E., Mannarino, A. P., Cohen, J. A., & Steer, R. A. (2006). A follow-up study of a multisite, randomized, controlled trial for children with sexual abuserelated PTSD symptoms. Journal of the American Academy of Child and Adolescent Psychiatry, 45(12), 1474e1484. Deblinger, E., Stauffer, L. B., & Steer, R. A. (2001). Comparative efficacies of supportive and cognitive behavioral group therapies for young children who
have been sexually abused and their nonoffending mothers. Child Maltreatment, 6(4), 332e343. Deblinger, E., Steer, R. A., & Lippmann, J. (1999). Two-year follow-up study of cognitive behavioral therapy for sexually abused children suffering post-traumatic stress symptoms. Child Abuse & Neglect, 23(12), 1371e1378. Feather, J. S., & Ronan, K. R. (2006). Trauma-focused cognitive-behavioural therapy for abused children with posttraumatic stress disorder: a pilot study. New Zealand Journal of Psychology, 35(3), 132e145. Giaconia, R. M., Reinherz, H. Z., Silverman, A. B., Pakiz, B., Frost, A. K., & Cohen, E. (1995). Traumas and posttraumatic stress disorder in a community population of older adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 34(10), 1369e1380. Giannopoulou, I., Dikaiakou, A., & Yule, W. (2006). Cognitive-behavioural group intervention for PTSD symptoms in children following the Athens 1999 earthquake: a pilot study. Clinical Child Psychology and Psychiatry, 11(4), 543e553. Hedges, L. V. (1981). Distribution theory for Glass’s estimator of effect sizes and related estimators. Journal of Statistics Education, 6(2), 107e128. Huedo-Medina, T., Sanchez-Meca, J., Marin-Martinez, F., & Botella, J. (2006). Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychological Methods, 11(2), 193e206. Jaberghaderi, N., Greenwald, R., Rubin, A., Zand, S. O., & Dalatabadi, S. (2004). A comparison of CBT and EMDR for sexually-abused Iranian girls. Clinical Psychology & Psychotherapy, 11(5), 358e368. Kazak, A. E., Alderfer, M. A., Streisand, R., Simms, S., Rourke, M. T., Barakat, L. P., et al. (2004). Treatment of posttraumatic stress symptoms in adolescent survivors of childhood cancer and their families: a randomized clinical trial. Journal of Family Psychology, 18(3), 493e504. Kendall, P. C. (1994). Treating anxiety disorders in children: results of a randomized clinical trial. Journal of Consulting and Clinical Psychology, 62(1), 100e110. Kessler, R. C., Sonnega, A., Bromet, E., Hughes, M., & Nelson, C. B. (1995). Posttraumatic stress disorder in the national comorbidity survey. Archives of General Psychiatry, 52(12), 1048e1060. King, N. J., Tonge, B. J., Mullen, P., Myerson, N., Heyne, D., Rollings, S., et al. (2000). Treating sexually abused children with posttraumatic stress symptoms: a randomized clinical trial. Journal of the American Academy of Child and Adolescent Psychiatry, 39(11), 1347e1355. Kolko, D. J. (1996). Individual cognitive behavioral treatment and family therapy for physically abused children and their offending parents: a comparison of clinical outcomes. Child Maltreatment, 1(4), 322e342. Leserman, J. (2005). Sexual abuse history: prevalence, health effects, mediators, and psychological treatment. Psychosomatic Medicine, 67(6), 906e915. Littell, J., Corcoran, J., & Pillai, V. (2008). Systematic reviews and meta-analysis. USA: Oxford University Press. MacDonald, G. M., Higgins, J. P. T., & Ramchandani, P. (2006). Cognitive-behavioural interventions for children who have been sexually abused (review). Cochrane Database of Systematic Reviews, 4. CD001930. March, J. S., Amaya-Jackson, L., Murray, M. C., & Schulte, A. (1998). Cognitivebehavioral psychotherapy for children and adolescents with posttraumatic stress disorder after a single-incident stressor. Journal of the American Academy of Child and Adolescent Psychiatry, 37(6), 585e593. Mueser, K. T., & Taub, J. (2008). Trauma and PTSD among adolescents with severe emotional disorders involved in multiple service systems. Psychiatric Services, 59(6), 627e634. Nikulina, V., Hergenrother, J. M., Brown, E. J., Doyle, M. E., Filton, B. J., & Carson, G. S. (2008). From efficacy to effectiveness: the trajectory of the
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413 treatment literature for children with PTSD. Expert Review of Neurotherapeutics, 8(8), 1233e1246. Robertson, M., Humphreys, L., & Ray, R. (2004). Psychological treatments for posttraumatic stress disorder: recommendations for the clinician. Journal of Psychiatric Practice, 10(2), 106e118. Ruggiero, K. J., & McLeer, S. V. (2000). PTSD scale of the child behavior checklist: concurrent and discriminant validity with non-clinic-referred sexually abused children. Journal of Traumatic Stress, 13(2), 287e299. Schwab-Stone, M. E., Ayers, T. S., Kasprow, W., Voyce, C., Barone, C., Shriver, T., et al. (1995). No safe haven: a study of violence exposure in an urban community. Journal of the American Academy of Child and Adolescent Psychiatry, 34(10), 1343e1352.
413
Smith, P., Yule, W., Perrin, S., Tranah, T., Dalgleish, T., & Clark, D. M. (2007). Cognitive-behavioral therapy for PTSD in children and adolescent: a preliminary randomized controlled trial. Journal of the American Academy of Child and Adolescent Psychiatry, 46(8), 1051e1061. Solomon, S. D., & Davidson, J. R. (1997). Trauma: prevalence, impairment, service use, and cost. Journal of Clinical Psychiatry, 58(9), 5e11. Stein, B. D., Jaycox, L. H., Katoaka, S. H., Wong, M., Tu, W., Elliott, M. N., et al. (2003). A mental health intervention for schoolchildren exposed to violence: a randomized controlled trial. Journal of the American Medical Association, 290 (5), 603e611. Wolfe, V. V., Gentile, C., & Wolfe, D. A. (1989). The impact of sexual abuse on children: a PTSD formulation. Behavior Therapy, 20(2), 215e228.
Contents lists available at ScienceDirect
Journal of Behavior Therapy and Experimental Psychiatry journal homepage: www.elsevier.com/locate/jbtep
Cognitive behavioral therapy for the treatment of pediatric posttraumatic stress disorder: A review and meta-analysis Joanna Kowalik a, *, Jennifer Weller a, Jacob Venter a, David Drachman b,1 a
University of Arizona College of Medicine Phoenix Campus, Maricopa Integrated Health System/District Medical Group, Desert Vista Behavioral Health Center, 570 West Brown Road, Mesa, AZ 85201, USA b Maricopa Integrated Health System, Maricopa Medical Center, 2601 E. Roosevelt Street, Phoenix, AZ 85008, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 August 2010 Received in revised form 18 January 2011 Accepted 2 February 2011
Background and objectives: There is no clear gold standard treatment for childhood posttraumatic stress disorder (PTSD). An annotated bibliography and meta-analysis were used to examine the efficacy of cognitive behavioral therapy (CBT) in the treatment of pediatric PTSD as measured by outcome data from the Child Behavior Checklist (CBCL). Method: A literature search produced 21 studies; of these, 10 utilized the CBCL but only eight were both 1) randomized; and 2) reported pre- and post-intervention scores. Results: The annotated bibliography revealed efficacy in general of CBT for pediatric PTSD. Using four indices of the CBCL, the meta-analysis identified statistically significant effect sizes for three of the four scales: Total Problems (TP; .327; p ¼ .003), Internalizing (INT; .314; p ¼ .001), and Externalizing (EXT; .192; p ¼ .040). The results for TP and INT were reliable as indicated by the fail-safe N and rank correlation tests. The effect size for the Total Competence (TCOMP; .054; p ¼ .620) index did not reach statistical significance. Limitations: Limitations included methodological inconsistencies across studies and lack of a randomized control group design, yielding few studies for meta-analysis. Conclusions: The efficacy of CBT in the treatment of pediatric PTSD was supported by the annotated bibliography and meta-analysis, contributing to best practices data. CBT addressed internalizing signs and symptoms (as measured by the CBCL) such as anxiety and depression more robustly than it did externalizing symptoms such as aggression and rule-breaking behavior, consistent with its purpose as a therapeutic intervention. Ó 2011 Elsevier Ltd. All rights reserved.
Keywords: PTSD CBT Pediatric Annotated bibliography Meta-analysis
1. Introduction The diagnosis of posttraumatic stress disorder (PTSD) first appeared in the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) in 1980 (American Psychiatric Association, 1980). Posttraumatic Stress Disorder is a complex disorder involving dysregulation of multiple neurobiological systems that affects cognitive, affective, and behavioral domains. Epidemiological studies report a prevalence rate of PTSD in the general adult population ranging between five and 14% (Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995). Trauma is frequently experienced in the United States, with an estimated one-
* Corresponding author. Tel.: þ1 480 344 2026; fax: þ1 480 344 0219. E-mail addresses: [email protected] (J. Kowalik), Jennifer_Weller@ dmgaz.org (J. Weller), [email protected] (J. Venter), David.Drachman@ mihs.org (D. Drachman). 1 Tel.: þ1 602 344 5161; fax: þ1 602 344 1974. 0005-7916/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jbtep.2011.02.002
third of the adult population exposed on an annual basis. Of these individuals, approximately 10e20% will develop the clinical syndrome of PTSD (Solomon & Davidson, 1997). The clinical impression is that, in contrast to the adult population, practitioners treating children and adolescents initially seemed reluctant to apply the diagnosis of PTSD to the pediatric age group. Data regarding trauma exposure as well as the subsequent development of PTSD in youth are more limited compared to adults. According to the National Child Traumatic Stress Network, 25% of children and adolescents experience a traumatic event by the time they reach 16 years of age (Copeland-Linder, 2008). A wide range of trauma rates has been reported in the literature, varying from as low as 16% (Cuffe et al., 1998) to as high as 40% in youth under the age of 18 (Boney-McCoy & Finklehor, 1996; Breslau, Davis, Andreski, & Peterson, 1991; Giaconia et al., 1995; SchwabeStone et al., 1995). The point prevalence of PTSD in youth remains unclear but lifetime prevalence estimates approximate 6% in the pediatric population (Giaconia et al., 1995). This number may be an
406
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
underestimate, given that greater numbers of youth are now receiving the diagnosis as clinicians’ awareness of and comfort with diagnosing PTSD increases. Despite clinicians’ slow recognition of PTSD in younger age groups, it is now accepted as a frequently occurring disorder. Posttraumatic stress disorder in children and adolescents is often severely disabling. Pediatric patients with PTSD present with a multitude of symptoms affecting functioning across different domains. Mueser and Taub (2008) reported a rate of PTSD as high as 20% among youth with severe emotional disorders who were involved in multiple systems of care. The authors also reported that adolescents with PTSD were more likely to engage in a variety of high-risk behaviors including running away from home, self-injury, and delinquency. Adolescents in their study reported higher levels of anxiety and depression and lower levels of optimal functioning in different settings (e.g., at home and at school) than did adolescents without PTSD. Different treatment approaches have been applied to address the symptoms of pediatric PTSD. Currently, outpatient psychotherapy is the preferred initial treatment modality for PTSD, with pharmacotherapy used as an adjunctive intervention (Cohen, 1998). Clinicians use Cognitive Behavioral Therapy (CBT) to address associations between stimuli and conditioned fear responses, the influence of environmental factors on symptom expression, and cognitive and affective regulation; therefore, CBT lends itself to the treatment of symptoms of PTSD. For treatment of sexually abused children, for example, clinicians often use CBT to address sequelae of the trauma including internalizing, externalizing, and sexualized behaviors (MacDonald, Higgins, & Ramchandani, 2006). Taking into account economic factors, CBT provides a focused, time-limited treatment approach to address the effects of trauma and is a cost-effective way of treating a larger number of individuals. Youth with PTSD often require a combination of treatment approaches (e.g., individual, group, and/or family psychotherapy along with pharmacotherapy) potentially in an array of treatment settings, with seamless transitions between levels of care. Treatment decisions are complicated by a lack of empirical data regarding outcomes of particular interventions. For these reasons, it remains difficult to recommend one particular treatment approach over another. Researchers have studied the use of psychotherapy in the pediatric PTSD population, with the majority of studies evaluating the efficacy of CBT approaches (Robertson, Humphreys, & Ray, 2004). Researchers have investigated trauma-focused CBT and found efficacy of this intervention in both individual and group therapy formats for sexually abused youth (Leserman, 2005). Pharmacological studies are fewer in number than studies of CBT, less rigorous in methodology and demonstrate less conclusive findings about efficacy and long-term outcomes (Nikulina et al., 2008). The original purpose of this article was to review the overall efficacy of CBT in the treatment of pediatric PTSD as described in recent literature. This review did not intend to evaluate the efficacy of combinations of treatment interventions or treatment delivered across different clinical settings. It examines immediate rather than long-term outcomes of the intervention. This review explores published research studies and contributes to the understanding and establishment of evidence-based treatment interventions. Once it was observed that the Child Behavior Checklist (CBCL) was the only measure utilized with some consistency across studies of pediatric PTSD, the more focused purpose of the study became to examine the efficacy of CBT in the treatment of pediatric PTSD as measured by outcome data from the CBCL. A quantitative approach that is well-suited for measuring the efficacy of interventions across multiple studies is meta-analysis. In this article, meta-analysis was used to evaluate outcomes from
randomized clinical trials of CBT in which the comparison was to an active control group. For inclusion in a meta-analysis, it is recommended that studies all utilize the same measure of the construct in question in order to maximize comparability among studies (Littell, Corcoran, & Pillai, 2008). Among the studies identified and reviewed in this article, the Child Behavior Checklist (CBCL) was the most commonly used outcome measure for treatment of pediatric PTSD (Achenbach & Edelbrock, 1983). The ability of the CBCL to distinguish between clinical and non-clinical patient samples has been well-established (Kendall, 1994). The CBCL was not designed to assess symptoms of PTSD in particular, and therefore is not considered a measure specific for PTSD phenomenology. It is a descriptive rating measure that assesses parent perceptions of their child’s behavior, adjustment, emotional functioning, and social functioning. Individual symptoms, but not the clinical syndrome of PTSD, are assessed by this measure; however, the CBCL provides composite indices that reflect how PTSD is expressed behaviorally by children and adolescents. The effects of trauma on children are varied and can be expressed in a number of ways. Symptoms can be categorized or conceptualized in ways that are similar to those used by the CBCL; these categories include internalizing, externalizing, and total competence. Some children who are exposed to traumatic experiences may react with internalizing symptoms (e.g., depression, anxiety, and/or somatic complaints as reflected by the Internalizing composite index of the CBCL), some with externalizing symptoms and behaviors (e.g., rulebreaking behaviors and/or aggression as assessed by the Externalizing composite index of the CBCL), and still others may manifest the effects of trauma with features that can be considered part of the “Total Competence” composite index of the CBCL (e.g., social challenges, diminished or limited participation in activities such as sports and hobbies, and school problems). These symptoms and signs are not specific to PTSD per se; however, the effects of trauma can be expressed in these different forms. 2. Method To evaluate the outcomes of CBT treatment studies of pediatric PTSD, the authors conducted a systematic search of data sources for relevant scientific publications. Articles were identified via a search of both Ovid MEDLINE and PsycINFO databases between 1966 and 2010. The following search terms were used: (PTSD OR posttraumatic stress disorder OR sexual abuse) AND (CBT OR cognitive behavioral therapy). The search was subsequently limited to the pediatric population (0e18 years) and the English language. The search identified 21 randomized controlled trials using CBT in the treatment of pediatric PTSD (Celano, Hazzard, Webb, & McCall, 1996; Cohen & Mannarino, 1996, 1998; Cohen, Deblinger, Mannarino, & Steer, 2004; Cohen, Mannarino, & Knudsen, 2005; Cohen, Mannarino, & Staron, 2006; Cohen, Mannarino, Perel, & Staron, 2007; Deblinger, Lippman, & Steer, 1996; Deblinger, Mannarino, Cohen, & Steer, 2006; Deblinger, Stauffer, & Steer, 2001; Deblinger, Steer, & Lippman, 1999; Feather & Ronan, 2006; Giannopoulou, Dikaiakou, & Yule, 2006; Jaberghaderi, Greenwald, Rubin, Zand, and Dalatabadi, 2004; Kazak et al., 2004; King et al., 2000; Kolko, 1996; March, Amaya-Jackson, Murray, & Schulte, 1998; Smith et al., 2007; Stein et al., 2003). Of the 21 studies, two were secondary analyses and therefore were excluded from further review. To be included in the meta-analysis, studies not only had to be randomized, but also had to use the same outcome measure before and immediately at conclusion of the intervention. Only studies comparing CBT to an active control group were included. Among the therapeutic approaches used in the active control groups were supportive unstructured psychotherapy, nondirective supportive treatment, and child-centered therapy. The CBCL was
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
407
Table 1 Annotated bibliography of studies of CBT in pediatric PTSD included in the meta-analysis. 1. Celano et al. (1996) in J Abn Child Psychology Objective Method Participants Interventions
Measures
Outcomes
Comments
To evaluate and compare the efficacy of two short-term individual therapy interventions for sexually abused girls and their non-offending female caretakers Randomized Controlled Trial Participants included 32 sexually abused girls aged 8e13 years and their caretakers from low-income African American families. Participants were randomly assigned to the experimental program (using CBT and metaphoric techniques) or control group (supportive unstructured psychotherapy). Both interventions consisted of eight 1-h sessions. In the experimental group, 30 min were spent with the child and 30 min with the caretaker per session. Two to three sessions included conjoint sessions. In the control group, 40e70% of the session time was with the child, 15e50% was with the caretaker, and 0e25% was with both parties. Out of 56 initial referrals, seven did not meet eligibility criteria, 17 dropped out, and 32 completed treatment. 1. CBCL: INT, EXT, PTSD subscale 2. Children’s Impact of Traumatic Events Scales e Revised (CITES-R) 3. Children’s Global Assessment Scale (CGAS) 4. Parent Reaction to Incest Disclosure Scale (PRIDS) 5. Parental Attribution Scale (PAS) Both treatment programs decreased PTSD symptoms and traumagenic beliefs reflecting self-blame and powerlessness, and increased overall psychosocial functioning. The experimental intervention was more effective than the comparison program in increasing abuse-related caretaker support of the child and in decreasing caretaker self-blame and expectations of undue negative impact of the abuse on the child. Length of the intervention was not specified
2. Cohen and Mannarino (1996) in J Am Acad Child Adolesc Psychiatry Objective Method Participants Interventions Measures Outcomes Comments
To assess treatment outcomes for sexually abused preschool-age children and their parents by comparing the effectiveness of CBT to nondirective supportive treatment (NST) Randomized Controlled Trial Participants included 67 sexually abused preschool children and their parents. Children were randomly assigned to either CBT adapted for sexually abused preschool children (CBT-SAP) or to nondirective supportive therapy. Child Measure: The Preschool Symptom Self-Report (PRESS) Parent Measures: CBCL-Parent Version, Child Sexual Behavior Inventory, Weekly Behavior Report Within-group comparison of pre- and post-treatment outcome measures demonstrated that the NST group did not change significantly with regard to symptomatology, but the CBT group showed significant improvement in symptoms. None
3. Deblinger et al. (1996) in Child Maltreatment Objective
To examine the differential effects of child and non-offending mother participation in CBT designed to treat PTSD and other behavioral and emotional difficulties in school-aged sexually abused children Method Randomized Controlled Trial Participants Participants included 100 families, of which 90 completed the pre-treatment and post-treatment assessment. Of the child group (aged 7e13 years), 83% were female, 17% were male, and 71% had a PTSD diagnosis. Interventions Children were randomly assigned to a control group or to one of three experimental treatment conditions: child only, mother only, or mother and child. Measures 1. Structured Background Interview (SBI) 2. Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS-E) 3. PTSD section of the K-SADS-E 4. CBCL 5. STAIC 6. CDI 7. Parenting Practice Questionnaire (PPQ) Outcomes Results of ANCOVAs showed significant main effects for the CBCL Externalizing scale, CDI, PTSD section of K-SADS-E, and PPQ. Children assigned to experimental treatment (child only or child and parent) reported fewer PTSD symptoms than children in the parent only or control groups. Treated children showed fewer externalizing behaviors. Only six of 38 children in the experimental group met PTSD criteria after the treatment. Comments Heterogeneity of community treatment; low sensitivity of instruments used to assess trauma 4. Cohen and Mannarino (1998) in Child Maltreatment Objective
To evaluate treatment outcomes for recently sexually abused children who received either sexual-specific CBT (SAS-CBT) or nondirective supportive therapy (NST). Method Randomized Controlled Trial Participants Participants included 82 children (7e14 years), but only 49 participants competed treatment and post-treatment evaluation. Interventions Children were randomly assigned to SAS-CBT or NST. Measures 1. CBCL 2. STAIC 3. CDI 4. CSBI Outcomes At post-treatment, children receiving SAS-CBT reported fewer depressive symptoms on the CDI than did children receiving NST. There was a group by time interaction on the CBCL Social Competence scale; the significant pre-treatment difference between groups suggested that the differential improvement of the SAS-CBT group represented regression to the mean. There was no significant difference between treatment groups at post-treatment evaluation, and no significant group by time interaction with regard to sexually inappropriate behaviors. Comments High drop-out rate 5. Deblinger, Steer, and Lippmann (1999) in Child Abuse and Neglect Objective Method
To determine if the 12 session pre- and post-intervention therapeutic gains found by Deblinger et al. (1996) were sustained two years after treatment Randomized Controlled Trial
408
Participants Interventions Measures
Outcomes
Comments
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
The 1996 sample included 100 participating families. Only 90 completed the pre- and post-treatment assessment. Of the children (aged 7e13), 83% were female, 17% were male, and 71% had a PTSD diagnosis. Children were randomly assigned to community comparison (control), a child-only treatment group, a mother-only treatment group, or a mother-and-child treatment group. Participants were assessed three, six, 12, and 24 months after treatment. 1. SBI 2. PTSD section of the K-SADS-E 3. CBCL 4. STAIC 5. CDI 6. PPQ A series of repeated MANCOVAs was conducted, controlling for pre-test scores. Results indicated that, for the measures of psychopathology on which symptoms decreased in the original study (i.e., externalizing problems, depression, and PTSD symptoms), scores on these measures at 3-month, 6-month, 1-year, and 2-year follow-ups were comparable to post-treatment scores. Incomplete data from measures at follow-ups
6. King et al. (2000) in J Am Acad Child Adolesc Psychiatry Objective Method Participants
To evaluate the efficacy of child and caregiver participation in the CBT treatment of sexually abused children with PTSD symptoms Randomized Controlled Trial Participants included 36 sexually abused boys and girls aged 5e17 years with symptoms meeting criteria for PTSD or considered “at high-risk” for PTSD. Only 28 children completed the study. Interventions Intervention groups included two experimental groups and one control group with 12 children in each group. Group 1 received child CBT consisting of 20 50-min sessions targeting abuse-related PTSD symptoms. Group 2 received family CBT consisting of 20 50-min sessions focused on parent training for behavior management and communication. Group 3 was a wait-list control group that received no contact for 24 weeks. An additional 10 children were screened but not included in the study. Eight more children dropped out during treatment (two from the control group, three from Group 1, and three from Group 2). Measures 1. From the Child Behavior Checklist (CBCL), the Internalizing and Externalizing composite scores and the PTSD subscale 2. Anxiety Disorders Interview Schedule (ADIS, Child Version), PTSD section 3. Children’s Depression Inventory (CDI) 4. Fear Thermometer for Sexually Abused Children (FT-SAC) 5. Revised Children’s Manifest Anxiety Scale (RCMAS) 6. Coping Questionnaire for Sexually Abused Children (CQ-SAC) 7. Global Assessment Functioning Scale (GAF) Outcomes Comparing Group 1 and Group 3, children receiving treatment reported significant improvement in PTSD symptoms and fear/anxiety. No difference was detected between child CBT and family CBT treatment groups. Results were still evident at 12-week follow-up. Comments Small study sample 7. Cohen et al. (2004) in J Am Acad Child Adolesc Psychiatry Objective
To examine the differential efficacy of trauma-focused cognitive behavioral therapy (TF-CBT) and child-centered therapy for treating PTSD and related emotional and behavioral problems in children who suffered sexual abuse Method Randomized Controlled Trial Participants Two hundred twenty-nine children (8e14 years) who experienced contact sexual abuse that was confirmed by CPS, law enforcement, or a professional independent evaluator were included in the study. Interventions Children were randomly assigned to the TF-CBT or the child-centered therapy group. Measures 1. Schedule for Affective Disorders and Schizophrenia for School-Aged Children e Present and Lifetime Version (K-SADS-PL) to assess DSM-IV psychiatric disorders 2. Children’s Depression Inventory (CDI) 3. State-Trait Inventory for Children (STAIC) 4. Children’s Attributions and Perceptions Scale (CAPS) Outcomes A series analyses of covariance indicated that children in TF-CBT, compared to those in child-centered therapy, demonstrated significantly more improvement with regard to PTSD, depression, behavior problems, shame, and abuse-related attributions. Comments Lack of non-treatment control group; small diversity of the sample group 8. Cohen et al. (2005) in Child Abuse and Neglect Objective Method Participants Interventions Measures
Outcomes Comments
To measure the durability of improvement in response to two treatments for sexually abused children Randomized Controlled Trial Participants included 82 sexually abused children (8e15 years) and their primary caretakers Children were randomly assigned to either TF-CBT or NST delivered over 12 sessions. 1. CDI 2. Trauma Symptom Checklist for Children (TSCC) 3. STAIC 4. CSBI 5. CBCL The intent-to-treat group indicated significant group time effects in favor of TF-CBT on measures of depression, anxiety, and sexual problems. Among treatment completers, the TF-CBT group showed greater improvement in anxiety, depression, sexual problems, and dissociation. Outcome measures lacked sensitivity for PTSD symptoms; high drop-out rate, especially in the NST group
the most commonly used outcome measure. Nine studies used a variety of scales as outcome measures other than the CBCL, and were excluded. Other outcome measures included but were not limited to the Weekly Behavior Report, the Post-Traumatic Stress Disorder Reaction Index, the State-Trait Anxiety Inventory for Children, the Revised Children’s Manifest Anxiety Scale, the Child Report of Post-Traumatic Symptoms, and the Subjective Units of
Distress Scale. The remaining studies of CBT utilized the CBCL; however, two did not include pre- and post-intervention assessments and were also excluded from the meta-analysis, leaving eight studies total for the meta-analysis. To obtain a comprehensive impression of the studies included in the meta-analysis, the authors compiled an annotated bibliography of each of the eight CBT studies (Table 1).
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
409
In an attempt to enable the CBCL to provide a more specific measure of PTSD symptoms, Wolfe, Gentile, and Wolfe (1989) derived a PTSD scale comprised of a subset of CBCL items. The authors illustrated that the PTSD scale has high discriminant validity, as sexually abused youth scored much higher than did a normative sample. The Wolfe PTSD scale was calculated in only two of the eight studies included in the meta-analysis. In the remaining six studies, scores on the set of items utilized to construct the PTSD scale were not available; therefore, scores for this scale could not be calculated for the remaining studies. The CBCL includes a number of composite indices, of which up to four were utilized to measure outcomes in the eight studies selected for meta-analysis: Total Problems (TP), Internalizing (INT), Externalizing (EXT), and Total Competence (TCOMP). Although the authors were not able to assess a measure specific for PTSD in the meta-analysis, the CBCL-based PTSD scale scores correlate very strongly with the CBCL TP, INT and EXT measures (Ruggiero & McLeer, 2000). Table 2 indicates which CBCL indices were reported in each study included in the meta-analysis. The purpose of the meta-analysis was to calculate an aggregate effect size for the CBT intervention for each of these four CBCL indices. The four metaanalyses were performed with the Comprehensive Meta-Analysis software application, Version 2.2.046 from Biostat, Inc., 2007.
2.3. Publication bias
2.1. Measure of effect size
3.1. Effect sizes
A commonly utilized measure of effect size is Cohen’s d which is simply the standardized difference between the mean outcomes in two experimental groups (Cohen & Yang, 2008). Because several studies included in this meta-analysis had small-sample sizes, Hedges’ g was used to correct for small-sample bias in the Cohen’s d measure and to express the effect size for each study (Hedges,1981). These effect sizes were then aggregated with each effect size weighted using the inverse variance method. This method provides effect sizes with smaller confidence intervals greater weight in the aggregated estimate of effect size (Littell et al., 2008). A negative effect size indicates a lower score on the CBCL index, suggesting improvement in the CBT group compared to the control group.
Fig. 1 and Table 3 report the effect sizes of each CBCL index. For the CBCL TP, INT and EXT indices, effect sizes were statistically significantly in favor of CBT over active control conditions. For the TCOMP index, the average effect size was not statistically significantly different across CBT and control groups. In other words, CBT interventions improved scores on the TP, INT and EXT indices relative to control groups but not on the TCOMP index. For all outcome measures assessed, the I2 statistic indicated substantial homogeneity among the eight studies. To verify the homogeneity assumption, both a fixed effects model (homogeneity assumed) and a random-effects model (heterogeneity assumed) were calculated. The results were exactly the same for TP, EXT, and TCOMP indices, and very similar for the INT index. These findings confirmed the appropriateness of using the fixed effects model in this study. Consequently, the fixed effects model was used to calculate the average effect sizes for each study included in the meta-analysis. (Figs. 2e5)
2.2. Heterogeneity To accurately estimate overall effect sizes for each CBCL index, it is important to know if the differences between calculated effect sizes for each study included in the meta-analysis can be explained by sampling error, or if the variation among effect sizes is due to systematic differences in study characteristics (e.g., such as differences in sample or treatment characteristics). The latter situation is referred to as heterogeneity. For each outcome index included in the meta-analysis, the I2 statistic was used to assess heterogeneity among the studies utilizing that specific measure (Huedo-Medina, Sanchez-Meca, Marin-Martinez, & Botella, 2006).
Studies without significant findings (i.e., null findings) are less likely to be submitted and/or accepted for publication than are studies that find statistically significant results. If unpublished, well-conducted studies with null findings were included in the meta-analysis, the effect size would be reduced. Therefore, it was important to determine if publication bias could have been a factor in the findings. One strategy for detecting publication bias in metaanalyses is to calculate the “fail-safe N.” The fail-safe N is the number of studies with null findings that would have to be added to the meta-analysis to reduce the effect size to a non-significant value (Littell et al., 2008). The rank correlation test can also be used to examine publication bias (Begg & Mazumdar, 1994). When publication bias exists, small studies are more likely to be included when they show a relatively large treatment effect, and more likely to be absent when they show a relatively small treatment effect. The result would then be an inverse correlation between study size and effect size. In this meta-analysis, both the fail-safe N and the rank correlation test were used to assess publication bias. 3. Results
3.2. Publication bias Table 2 reports the fail-safe N for each measure. The fail-safe N for both the TP and INT analyses was four studies. Therefore, it is
Table 2 CBCL Indices reported in studies included in both the annotated bibliography and meta-analysis. Authors Celano et al. (1996) Cohen and Mannarino (1996) Deblinger et al. (1996) Cohen and Mannarino (1998) Deblinger et al. (1999) King et al. (2000) Cohen et al. (2004) Cohen et al. (2005) TOTAL
TP
INT
EXT
X
X X X X
X X 4
X X X 7
X X X X X X X X 8
X
TCOMP
PTSD X
X X X X X 4
2 Fig. 1. Average effect sizes and confidence intervals.
410
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
Table 3 Results of meta-analysis of each CBCL index. Index
Number of studies
Confidence interval
Average effect size
Z statistic
p value
I2
Fail-safe N
Rank correlation
TP INT EXT TCOMP
4 7 8 4
.541 .505 .376 .267
.327 .314 .192 .054
2.297 3.550 2.051 .496
.003 .001 .040 .620
0.0% 20.3% 0.0% 0.0%
4 4 0 0
.167a .571a .179a .167a
a
to to to to
.113 .122 .008 .159
Test for a negative correlation was not significant.
unlikely that publication bias affected the findings for these two CBCL indices. On the other hand, the fail-safe N for the EXT and TCOMP indices was so small (zero) that findings on these two indices should be viewed with caution. None of the rank correlation tests for each of the four CBCL outcome indices showed a significant inverse correlation. These results provided further evidence that publication bias did not significantly affect the findings. 4. Conclusions 4.1. Study strengths and implications The annotated bibliography in Table 1 supported the efficacy in general of CBT for treatment of pediatric PTSD. The meta-analysis of eight randomized trials of CBT, comparing CBT to active control groups and using both pre- and post-intervention assessments, provided strong evidence that CBT is effective in the treatment of childhood PTSD. In particular, Total Problems, Internalizing, and Externalizing indices of the CBCL showed favorable outcomes as reflected by greater effect sizes of the CBT treatment groups versus comparison groups. The findings for the effects of CBT on the Total Competence index of the CBCL were not statistically significant. Participants in the comparison groups in the studies included in this meta-analysis received active treatment as opposed to no treatment at all. This fact strengthens the significance of our conclusion that CBT is efficacious for the treatment of pediatric PTSD. The fail-safe N for both the Total Problems and Internalizing indices was four studies. Therefore, despite the small number of studies available for inclusion in the meta-analysis, the authors can be confident that neither publication bias nor the small number of studies suitable for inclusion significantly affected the findings reflected by these two CBCL indices. In contrast, the fail-safe N for
the Externalizing and Total Competence indices was zero, casting doubt on the validity of effect sizes observed for these dimensions of functioning. Results for the Total Problems and Internalizing indices provided support for the effectiveness of CBT in addressing symptoms specifically measured by these two CBCL indices. Statistically significant positive change occurred in these two indices in response to CBT, and the fail-safe N for these indices indicated that the findings were reliable. Although the Externalizing index showed statistically significant improvement in response to CBT, the fail-safe N of zero indicated that the results should be interpreted with caution. The Total Competence index did not change following CBT and had a fail-safe N of zero; these findings could not be explained satisfactorily and require further exploration. It is possible that CBT interventions for pediatric PTSD do not address externalizing and Total Competence constructs or behaviors as measured on the CBCL to the same degree as they do internalizing symptoms. It would seem that CBT better addresses internalizing symptoms such as anxiety and depression (as the intervention was originally intended to do) than it does externalizing symptoms such as aggression and/or rule-breaking behaviors as they are measured by the CBCL. Externalizing symptoms often lead parents to seek treatment for their children with PTSD, due to significant effects of aggression and rule-breaking behavior on overall social functioning. In addition to CBT, other treatment approaches may be helpful specifically targeting externalizing signs and symptoms that manifest in response to PTSD. 4.2. Study limitations For inclusion in a meta-analysis, studies must utilize similar (or ideally, identical) outcome measures and must use a randomized control group design. The current meta-analysis was limited by
Most Recent
Earliest
Cohen & Mannarino Jan 1996*
Cohen & Mannarino Feb 1998
Each line represents the effect size for the indicated study and all studies occurring before it
Cohen et al. Apr 2004* Cohen et al. 2005*
-1.0
Effect Size
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
Average Effect Size
Favors CBT Group
95% Confidence Interval
Favors Control Group
* Statistically significant
Fig. 2. Cumulative evidence chart: CBCL total problems (TP).
0.8
1.0
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
411
Earliest
Celano et al. Jan 1996 Cohen & Mannarino Jan 1996* Deblinger et al. Nov 1996
Each line represents the effect size for the indicated study and all studies occurring before it
Most Recent
Cohen & Mannarino Feb 1998* King et al. Nov 2000 Cohen et al. Apr 2004* Cohen et al. 2005* -1.0
-0.8
-0.6
-0.4
Effect Size
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Average Effect Size
Favors CBT Group
Favors Control Group
* Statistically significant
95% Confidence Interval
Fig. 3. Cumulative evidence chart: CBCL internalizing (INT).
the diversity of outcome measures utilized across studies, and the absence of randomized control group designs. Some of the outcome measures were designed specifically to measure childhood traumatic experiences, while others tapped emotional and/ or behavioral problems that may be a result of trauma more generally. The CBCL was used most often as an outcome measure among CBT treatment studies; therefore, only studies that utilized this measure were included in the meta-analysis. The use of diverse measures across studies revealed the lack of a “gold standard” instrument for use in the assessment of pediatric PTSD. With respect to use of a randomized control group design, most studies of the treatment of pediatric PTSD using CBT approaches did not utilize such a research design. A further limitation due to the relatively low number of studies that could be included in the meta-analysis was that it limited the generalizability of the findings. A greater number of studies of CBT utilizing the same outcome measure(s), and that were more specifically aimed at assessing PTSD symptoms, would have yielded more robust findings. Inconsistencies across studies in methodology further limited the results of the meta-analysis. Variables such as the type and fidelity of CBT used to address pediatric PTSD (e.g., trauma-focused
CBT, combining traditional CBT with metaphorical techniques, etc.), participant characteristics (e.g., child only receiving intervention versus child and parent receiving intervention), therapist sophistication or training in the intervention(s), and number of sessions and the length of the treatment intervention period all could have affected study findings. Additional clinical characteristics such as age, sex differences, sample sizes, differences in drop-out rates, and severity of symptoms in the study samples could affect the generalizability of findings. 4.3. Future directions Given that this review provided support for the use of CBT to treat symptoms of pediatric PTSD, a next step would be to deconstruct the components of CBT to determine which symptoms of PTSD respond best to particular aspects of this treatment approach. Doing so would allow clinicians to select the ideal combination of treatment components for their pediatric patients who have been exposed to traumatic experiences. Additionally, future studies of the use of CBT in pediatric PTSD would benefit from use of a randomized control group design to enhance scientific merit.
Earliest
Celano et al. Jan 1996 Cohen & Mannarino Jan 1996* Deblinger et al. Nov 1996*
Each line represents the effect size for the indicated study and all studies occurring before it
Most Recent
Cohen & Mannarino Feb 1998 Deblinger et al. 1999 King et al. Nov 2000 Cohen et al. Apr 2004* Cohen et al. 2005* -1.5
Effect Size
-1.0
-0.5
0.0
0.5
Average Effect Size
Favors CBT Group
95% Confidence Interval
Favors Control Group
* Statistically significant
Fig. 4. Cumulative evidence chart: CBCL externalizing (EXT).
1.0
1.5
412
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413
Most Recent
Earliest
Cohen & Mannarino Jan 1996
Cohen & Mannarino Feb 1998
Each line represents the effect size for the indicated study and all studies occurring before it
Cohen et al. Apr 2004
Cohen et al. 2005
-1.0
-0.8
-0.6
Effect Size
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Average Effect Size
Favors CBT Group
95% Confidence Interval
Favors Control Group
* Statistically significant
Fig. 5. Cumulative evidence chart: CBCL total competence (TCOMP).
References American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, D.C.: American Psychiatric Press. (DSM-III). Achenbach, T., & Edelbrock, C. (1983). Manual for the child behavior checklist and revised child behavior profile. Univ Vermont/Dept Psychiatry. Begg, C. B., & Mazumdar, M. (1994). Operating characteristics of a rank correlation test for publication bias. Biometrics, 50(4), 1088e1101. Boney-McCoy, S., & Finklehor, D. (1996). Is youth victimization related to trauma symptoms and depression after controlling for prior symptoms and family relationships? A longitudinal, prospective study. Journal of Consulting and Clinical Psychology, 64(6), 1406e1416. Breslau, N., Davis, G. C., Andreski, P., & Peterson, E. (1991). Traumatic events and postrraumatic stress disorder in an urban population of young adults. Archives of General Psychiatry, 48(3), 216e222. Celano, M., Hazzard, A., Webb, C., & McCall, C. (1996). Treatment of traumagenic beliefs among sexually abused girls and their mothers: an evaluation study. Journal of Abnormal Child Psychology, 24(1), 1e17. Cohen, J. A. (1998). Practice parameters for the assessment and treatment of children and adolescents with posttraumatic stress disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 37(10 Suppl.), 4se26s. Cohen, J. A., Deblinger, E., Mannarino, A. P., & Steer, R. (2004). A multisite, randomized controlled trial for children with sexual abuse-related PTSD symptoms. Journal of the American Academy of Child and Adolescent Psychiatry, 43(4), 393e402. Cohen, J. A., & Mannarino, A. P. (1996). A treatment outcome study for sexually abused preschool children: initial findings. Journal of the American Academy of Child and Adolescent Psychiatry, 35(1), 42e50. Cohen, J. A., & Mannarino, A. P. (1998). Interventions for sexually abused children: initial treatment outcome findings. Child Maltreatment, 3(1), 17e26. Cohen, J. A., Mannarino, A. P., & Knudsen, K. (2005). Treating sexually abused children: 1 year follow-up of a randomized controlled trial. Child Abuse and Neglect, 29(2), 103e105. Cohen, J. A., Mannarino, A. P., Perel, J. M., & Staron, V. (2007). A pilot randomized controlled trial of combined trauma-focused CBT and sertraline for childhood PTSD symptoms. Journal of the American Academy of Child and Adolescent Psychiatry, 46(7), 811e819. Cohen, J. A., Mannarino, A. P., & Staron, V. R. (2006). A pilot study of modified cognitive-behavioral therapy for childhood traumatic grief (CBT-CTG). Journal of the American Academy of Child and Adolescent Psychiatry, 45(12), 1465e1473. Cohen, J., & Yang, F. (2008). Progress in applied mathematical modeling. Nova Science Pub Inc. Copeland-Linder, N. (2008). Posttraumatic stress disorder. Pediatrics in Review, 29, 103e104. Cuffe, S. P., Addy, C. L., Garrison, C. Z., Waller, J. L., Jackson, K. L., McKeown, R. E., et al. (1998). Prevalence of PTSD in a community sample of older adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 37(2), 147e154. Deblinger, E., Lippman, J., & Steer, R. (1996). Sexually abused children suffering posttraumatic stress symptoms: initial treatment outcome findings. Child Maltreatment, 1(4), 310e321. Deblinger, E., Mannarino, A. P., Cohen, J. A., & Steer, R. A. (2006). A follow-up study of a multisite, randomized, controlled trial for children with sexual abuserelated PTSD symptoms. Journal of the American Academy of Child and Adolescent Psychiatry, 45(12), 1474e1484. Deblinger, E., Stauffer, L. B., & Steer, R. A. (2001). Comparative efficacies of supportive and cognitive behavioral group therapies for young children who
have been sexually abused and their nonoffending mothers. Child Maltreatment, 6(4), 332e343. Deblinger, E., Steer, R. A., & Lippmann, J. (1999). Two-year follow-up study of cognitive behavioral therapy for sexually abused children suffering post-traumatic stress symptoms. Child Abuse & Neglect, 23(12), 1371e1378. Feather, J. S., & Ronan, K. R. (2006). Trauma-focused cognitive-behavioural therapy for abused children with posttraumatic stress disorder: a pilot study. New Zealand Journal of Psychology, 35(3), 132e145. Giaconia, R. M., Reinherz, H. Z., Silverman, A. B., Pakiz, B., Frost, A. K., & Cohen, E. (1995). Traumas and posttraumatic stress disorder in a community population of older adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 34(10), 1369e1380. Giannopoulou, I., Dikaiakou, A., & Yule, W. (2006). Cognitive-behavioural group intervention for PTSD symptoms in children following the Athens 1999 earthquake: a pilot study. Clinical Child Psychology and Psychiatry, 11(4), 543e553. Hedges, L. V. (1981). Distribution theory for Glass’s estimator of effect sizes and related estimators. Journal of Statistics Education, 6(2), 107e128. Huedo-Medina, T., Sanchez-Meca, J., Marin-Martinez, F., & Botella, J. (2006). Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychological Methods, 11(2), 193e206. Jaberghaderi, N., Greenwald, R., Rubin, A., Zand, S. O., & Dalatabadi, S. (2004). A comparison of CBT and EMDR for sexually-abused Iranian girls. Clinical Psychology & Psychotherapy, 11(5), 358e368. Kazak, A. E., Alderfer, M. A., Streisand, R., Simms, S., Rourke, M. T., Barakat, L. P., et al. (2004). Treatment of posttraumatic stress symptoms in adolescent survivors of childhood cancer and their families: a randomized clinical trial. Journal of Family Psychology, 18(3), 493e504. Kendall, P. C. (1994). Treating anxiety disorders in children: results of a randomized clinical trial. Journal of Consulting and Clinical Psychology, 62(1), 100e110. Kessler, R. C., Sonnega, A., Bromet, E., Hughes, M., & Nelson, C. B. (1995). Posttraumatic stress disorder in the national comorbidity survey. Archives of General Psychiatry, 52(12), 1048e1060. King, N. J., Tonge, B. J., Mullen, P., Myerson, N., Heyne, D., Rollings, S., et al. (2000). Treating sexually abused children with posttraumatic stress symptoms: a randomized clinical trial. Journal of the American Academy of Child and Adolescent Psychiatry, 39(11), 1347e1355. Kolko, D. J. (1996). Individual cognitive behavioral treatment and family therapy for physically abused children and their offending parents: a comparison of clinical outcomes. Child Maltreatment, 1(4), 322e342. Leserman, J. (2005). Sexual abuse history: prevalence, health effects, mediators, and psychological treatment. Psychosomatic Medicine, 67(6), 906e915. Littell, J., Corcoran, J., & Pillai, V. (2008). Systematic reviews and meta-analysis. USA: Oxford University Press. MacDonald, G. M., Higgins, J. P. T., & Ramchandani, P. (2006). Cognitive-behavioural interventions for children who have been sexually abused (review). Cochrane Database of Systematic Reviews, 4. CD001930. March, J. S., Amaya-Jackson, L., Murray, M. C., & Schulte, A. (1998). Cognitivebehavioral psychotherapy for children and adolescents with posttraumatic stress disorder after a single-incident stressor. Journal of the American Academy of Child and Adolescent Psychiatry, 37(6), 585e593. Mueser, K. T., & Taub, J. (2008). Trauma and PTSD among adolescents with severe emotional disorders involved in multiple service systems. Psychiatric Services, 59(6), 627e634. Nikulina, V., Hergenrother, J. M., Brown, E. J., Doyle, M. E., Filton, B. J., & Carson, G. S. (2008). From efficacy to effectiveness: the trajectory of the
J. Kowalik et al. / J. Behav. Ther. & Exp. Psychiat. 42 (2011) 405e413 treatment literature for children with PTSD. Expert Review of Neurotherapeutics, 8(8), 1233e1246. Robertson, M., Humphreys, L., & Ray, R. (2004). Psychological treatments for posttraumatic stress disorder: recommendations for the clinician. Journal of Psychiatric Practice, 10(2), 106e118. Ruggiero, K. J., & McLeer, S. V. (2000). PTSD scale of the child behavior checklist: concurrent and discriminant validity with non-clinic-referred sexually abused children. Journal of Traumatic Stress, 13(2), 287e299. Schwab-Stone, M. E., Ayers, T. S., Kasprow, W., Voyce, C., Barone, C., Shriver, T., et al. (1995). No safe haven: a study of violence exposure in an urban community. Journal of the American Academy of Child and Adolescent Psychiatry, 34(10), 1343e1352.
413
Smith, P., Yule, W., Perrin, S., Tranah, T., Dalgleish, T., & Clark, D. M. (2007). Cognitive-behavioral therapy for PTSD in children and adolescent: a preliminary randomized controlled trial. Journal of the American Academy of Child and Adolescent Psychiatry, 46(8), 1051e1061. Solomon, S. D., & Davidson, J. R. (1997). Trauma: prevalence, impairment, service use, and cost. Journal of Clinical Psychiatry, 58(9), 5e11. Stein, B. D., Jaycox, L. H., Katoaka, S. H., Wong, M., Tu, W., Elliott, M. N., et al. (2003). A mental health intervention for schoolchildren exposed to violence: a randomized controlled trial. Journal of the American Medical Association, 290 (5), 603e611. Wolfe, V. V., Gentile, C., & Wolfe, D. A. (1989). The impact of sexual abuse on children: a PTSD formulation. Behavior Therapy, 20(2), 215e228.